Genetic Algorithm
A gradient-free optimization algorithm proposed by John Holland in the 1960s, inspired by biological evolution. It searches for optimal solutions through mechanisms of mutation, recombination, and selection.
Timeline (last 90 days)
The study selected four hyperparameter optimization strategies: genetic algorithm, Hyperband, Optuna, and SMAC3
Evolutionary search is a class of black-box optimization algorithms inspired by biological evolution, with typical representatives including genetic algorithms (GA), evolution strategies (ES), and CMA-ES.
A developer attempted to build a 2048 game auto-solver using a genetic algorithm combined with Expectimax search; the current implementation can reach the 2048 tile in 30% to 50% of games.
The role of genetic algorithms in 2048 solvers is typically to optimize the weights of each feature in the Expectimax evaluation function.
If features are not normalized, features with larger magnitudes will dominate the evaluation function, rendering the weights optimized by the genetic algorithm meaningless.
The recommended progression is: first use ablation experiments to confirm feature effectiveness, then implement 64-bit integer encoding for acceleration, then gradually increase search depth, and finally use genetic algorithms to re-optimize feature weights at the new depth.
The schema theorem in genetic algorithms states that short, high-fitness building blocks (schemata) spread exponentially through the population.
Modern VRP solution methods typically combine metaheuristic algorithms (such as genetic algorithms and simulated annealing) with machine learning prediction models to approximate the optimal solution within an acceptable timeframe.
A developer combined a genetic algorithm with an MLP neural network to allow a population of 3D robotic arms to autonomously evolve the ability to reach a target.
Evolutionary computation methods such as genetic algorithms do not require gradient information, making them suitable for black-box optimization problems where the physics simulator is non-differentiable or rewards are sparse.
1 more timeline events
All Facts (10)
John Holland proposed the theoretical framework of genetic algorithms in the 1960s.
60%UnverifiedThe study selected four hyperparameter optimization strategies: genetic algorithm, Hyperband, Optuna, and SMAC3
50%UnverifiedEvolutionary search is a class of black-box optimization algorithms inspired by biological evolution, with typical representatives including genetic algorithms (GA), evolution strategies (ES), and CMA-ES.
50%UnverifiedThe role of genetic algorithms in 2048 solvers is typically to optimize the weights of each feature in the Expectimax evaluation function.
50%UnverifiedIf features are not normalized, features with larger magnitudes will dominate the evaluation function, rendering the weights optimized by the genetic algorithm meaningless.
50%UnverifiedThe recommended progression is: first use ablation experiments to confirm feature effectiveness, then implement 64-bit integer encoding for acceleration, then gradually increase search depth, and finally use genetic algorithms to re-optimize feature weights at the new depth.
50%UnverifiedThe schema theorem in genetic algorithms states that short, high-fitness building blocks (schemata) spread exponentially through the population.
50%UnverifiedModern VRP solution methods typically combine metaheuristic algorithms (such as genetic algorithms and simulated annealing) with machine learning prediction models to approximate the optimal solution within an acceptable timeframe.
50%UnverifiedEvolutionary computation methods such as genetic algorithms do not require gradient information, making them suitable for black-box optimization problems where the physics simulator is non-differentiable or rewards are sparse.
50%Unverified爬山算法存在陷入局部最优的风险，研究者发展出模拟退火、随机重启爬山、遗传算法等变体来跳出局部最优
50%